Hydrosight

HydroSight AI converts crowdsourced street-level flood photos into GIS-ready 3D flood-extent shapefiles, building inundation depth profiles, and safe emergency rescue routes for flood-prone Rodriguez, Rizal. Using Meta's sam-3.1 and muse-spark-1.3, it measures water depth from submerged reference objects with an autonomous confidence self-check loop. The engine then ray-casts the camera viewcone onto a 5-meter bare-earth DEM (EPSG:32651), floods connected lowlands across 83,000+ buildings, clips impassable roads on the street network, and routes rescue vehicles to verified dry evacuation centers. All code was written during the hackathon period.